[zeromq-dev] Crash in zeromq 3.2.4 Windows x64

Robin Scher robin at uberware.net
Fri Feb 14 03:56:13 CET 2014


Hi List.

I’ve been using ZeroMQ for about 8 months and it has been working well for my project. Recently I’ve noticed a crash on a customer site running under moderately heavy use (this machine is communicating with about 850 active clients, but the crash seems to happen over time, not based on how many clients are connected at the time it happens). I didn’t get a full debug report because it happened when I wasn’t directly monitoring the system and the process had to be restarted because it is deployed in production. However, I have the call stack where it happened:

Unhandled exception at 0x000007FEFD0BBCCD (KernelBase.dll) in SmedgeMaster.exe: 0x40000015:  %hs (parameters: 0x00000000004EF078).

        KernelBase.dll!000007fefd0bbccd()       Unknown
>       SmedgeLib.dll!zmq::tune_tcp_socket(unsigned __int64)    Unknown
        SmedgeLib.dll!zmq::tcp_connecter_t::out_event(void)     Unknown
        SmedgeLib.dll!zmq::select_t::loop(void) Unknown
        SmedgeLib.dll![thunk]:zmq::tcp_connecter_t::`vector deleting destructor'`adjustor{504}' (unsigned int)  Unknown
        msvcr110.dll!_callthreadstartex() Line 354      C
        msvcr110.dll!_threadstartex(void * ptd) Line 332        C
        kernel32.dll!0000000076d5652d() Unknown
        ntdll.dll!0000000076e8c521()    Unknown

The code is compiled Windows 64 bit on MSVC 11 (Visual Studio 2012). This is an optimized release build with debugging info. I build zmq as a static library that is linked into my own dynamic library. The thread in question is one of the threads that is started by zmq itself, not my own library or code. I’m afraid I didn’t get the exact instruction in the zmq code that triggered the exception before I had to restart it (there was some panic as the process was in the middle of a lot of work and clients were waiting!)

Does anyone have any idea what may be going on here? Is there some way I can catch whatever is failing so I can handle it more gracefully than just killing my entire process and potentially losing data? I’ve been having crashing like this every few days of operation, but this is the first time I was even able to get it to happen while the debugger was attached. I have the debugger attached again after a restart, so if it happens again I may be able to get more information, especially if someone can help me to know what I should be looking for exactly.

Thank you for any insight you can provide. 
-robin

Robin Scher
robin at uberware.net
+1 (213) 448-0443



-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 496 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <https://lists.zeromq.org/pipermail/zeromq-dev/attachments/20140213/a8394468/attachment.sig>


More information about the zeromq-dev mailing list